Abstract
Integrating with the signal processing, inverse Radon transform, and the variational model, the framework at least saving 83% data acquisition time for fast, smooth three-dimensional (3D) reconstruction from the limited dataset is elucidated in the field of terahertz imaging applications. In hot pursuit, under the viewpoint of discrete geometry, the quantifiable comparison for 3D surfaces by computing the standard deviation of mean curvatures is also proposed to show the reconstructed effectiveness from less input with gaps.